The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes powerful CLI tools including gh, glab, and git to modify branch protection settings and rewrite repository history via force-push operations. These are high-risk commands intended for administrative tasks.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it processes untrusted data from remote repository APIs. Ingestion points: API responses from gh api graphql and glab api containing branch names and protection rule details. Boundary markers: Mandatory explicit user confirmation is required for all write operations; instructions state that 'Workflow auto-advance is not acceptance'. Capability inventory: Capability to create, update, and delete protection rules, and execute git push --force-with-lease (documented in references/github.md and references/gitlab.md). Sanitization: The skill explicitly requires URL-encoding for branch names in GitLab API calls to prevent path traversal or command issues.

git-branch-protect